Firecrawl Proxy Setup with Mobile IPs in 2026

version: "3.9"
services:
  firecrawl-api:
    image: firecrawl/firecrawl:v2.3.1
    ports:
      - "3002:3002"
    environment:
      # --- Coronium 4G BYOP ---
      PROXY_SERVER: "https://proxy.coronium.io:8000"
      PROXY_USERNAME: "${CORONIUM_USER}"
      PROXY_PASSWORD: "${CORONIUM_PASS}"
      PROXY_ROTATE_URL: "https://coronium.io/api/rotate?key=${CORONIUM_KEY}"
      # --- Firecrawl core ---
      REDIS_URL: "redis://redis:6379"
      PLAYWRIGHT_MICROSERVICE_URL: "http://playwright:3003"
      USE_DB_AUTHENTICATION: "false"
      TEST_API_KEY: "fc-local-dev-key"
    depends_on: [redis, playwright]

  playwright:
    image: firecrawl/playwright-service:latest
    environment:
      PROXY_SERVER: "${PROXY_SERVER}"
      PROXY_USERNAME: "${PROXY_USERNAME}"
      PROXY_PASSWORD: "${PROXY_PASSWORD}"
    ports:
      - "3003:3003"

  redis:
    image: redis:7-alpine
    ports: ["6379:6379"]

from firecrawl import FirecrawlApp
import os, requests

# Self-hosted Firecrawl already configured with Coronium via docker-compose.
app = FirecrawlApp(
    api_key="fc-local-dev-key",
    api_url="http://localhost:3002",
)

# Rotate IP before a sensitive target
requests.get(f"https://coronium.io/api/rotate?key={os.environ['CORONIUM_KEY']}")

result = app.scrape_url(
    "https://www.instagram.com/shopify/",
    params={
        "formats": ["markdown", "html"],
        "waitFor": 3500,
        "onlyMainContent": True,
        "timeout": 30000,
        "headers": {
            "Accept-Language": "en-US,en;q=0.9",
            "X-Coronium-Session": "shopify-scrape-042026",
        },
    },
)

print(f"Success: {result['success']}")
print(f"Credits used: {result['metadata']['creditsUsed']}")
print(result["markdown"][:500])

# Kick off a full-site crawl, 500 pages, rotate every 25
curl -X POST https://api.firecrawl.dev/v2/crawl \
  -H "Authorization: Bearer $FIRECRAWL_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "url": "https://shop.example.com",
    "limit": 500,
    "scrapeOptions": {
      "formats": ["markdown"],
      "onlyMainContent": true,
      "headers": {
        "X-Proxy-Pool": "coronium-4g-us-east",
        "X-Rotate-Every": "25"
      }
    },
    "includePaths": ["/products/*", "/collections/*"],
    "excludePaths": ["/account/*", "/cart"],
    "maxDepth": 5
  }'

# Poll the crawl job
curl https://api.firecrawl.dev/v2/crawl/$JOB_ID \
  -H "Authorization: Bearer $FIRECRAWL_KEY"

from llama_index.readers.web import FirecrawlReader
from llama_index.core import VectorStoreIndex, Settings
from llama_index.embeddings.openai import OpenAIEmbedding
from llama_index.llms.anthropic import Anthropic

Settings.llm = Anthropic(model="claude-opus-4-7")
Settings.embed_model = OpenAIEmbedding(model="text-embedding-3-large")

reader = FirecrawlReader(
    api_key="fc-local-dev-key",
    api_url="http://localhost:3002",     # Coronium-backed self-hosted
    mode="crawl",
)

documents = reader.load_data(
    url="https://www.shopify.com/partners/blog",
    params={
        "limit": 300,
        "scrapeOptions": {
            "formats": ["markdown"],
            "onlyMainContent": True,
            "headers": {"X-Proxy-Pool": "coronium-4g-us-east"},
        },
    },
)

index = VectorStoreIndex.from_documents(documents)
query_engine = index.as_query_engine(similarity_top_k=6)

answer = query_engine.query(
    "Summarise the three most impactful Shopify partner launches in Q1 2026."
)
print(answer)
for node in answer.source_nodes:
    print("Source:", node.metadata.get("source"))